System overview

genomeOS separates the journey into six parts. Contributors use the codes P0 through P5 as short project-board labels; each code is translated below before its technical role is described. The information moves in one direction:

original sources → population locations → measured evidence → frequency models
→ disease-burden models → versioned result files → data service → interactive map

Population locations (internal label: P0)

The population-location registry records where an independently sampled population can be located, how uncertain that location is, and where the location and community-governance information came from. If a source does not support a location or uncertainty range, the build stops instead of guessing one.

Measured evidence (internal label: P1)

The observation store preserves what a source measured: the allele being counted, its count and denominator, the sampled population, how participants were recruited, the study group, laboratory method, citation, and exact table or figure location. Proposed rows remain in a review area until every required field is independently verified and the population has a supported location.

Frequency and disease-burden modeling (internal labels: P2 and P3)

Frequency models estimate how a variant may vary between measured locations and report both uncertainty and places with inadequate evidence. Disease-burden models combine supported frequency results with inheritance, the chance that a variant produces a condition, and population counts. Both run before results reach the website. Missing evidence or required assumptions produce no number.

Data service and interactive map (internal labels: P4 and P5)

The data service reads finished, versioned result files; it never fits a model while someone is using the website. Today the interactive map reads a static, versioned export of those files rather than calling the service. It shows measured evidence, modeled patterns, uncertainty and insufficient-evidence areas as clearly separate layers; disease-burden estimates will be added as a further layer. Every shareable view identifies the model and data versions behind it.

See the scientific and engineering objectives for acceptance evidence at each boundary.